Dynamic Reliability Assessment for Nonrepairable Multistate Systems by Aggregating Multilevel Imperfect Inspection Data

Traditional time-based reliability assessment methods compute reliability measures of a multistate system (MSS) purely based upon historical time-to-failure data collected from a large population of identical systems. Using these methods, one can only assess the reliability of a system from a population or statistical perspective. Moreover, these methods fail to characterize the stochastic behavior of a specific individual MSS over time. Accordingly, in this paper, a dynamic reliability assessment method that can aggregate inspection data across multiple levels (such as component level, subsystem level, and system level) of a nonrepairable MSS has been studied. In general, inspection data collected from multiple levels of a system can be imperfect, but they are stochastically correlated with the actual states of the inspected system and components. A set of two-stage recursive Bayesian formulations has been put forth to dynamically update the reliability function of a specific MSS over time by utilizing imperfect inspection data collected simultaneously or asynchronously from multiple levels of the system. The proposed method is exemplified via an illustrative example of an underground flow transmission system. The impact of the probability of detection on the accuracy of the remaining useful life prediction is also examined.